Integrated tunable RF notch filter
Large interfering signals (interferers) with spectra near a desired signal can cause distortion in a wireless receiver due to a non-linear signal path. It is typically a performance advantage to attenuate these interferers earlier in the signal path, rather than later in the signal path, because these interferers can cause saturation of amplifying stages. In certain situations, the frequency offset of an interfering signal, with respect to the desired signal, can be on the order of 10 megahertz (MHz), whereas the center frequencies can be on the order of several gigahertz (GHz). Thus, a filter with “baseband” precision would be needed at radio frequency to notch out the interferer, which is relatively difficult to do. Disclosed is a technique to estimate the relative strength and center frequency of the interferer and to place the center frequency of a notch filter adaptively and precisely at the interferer location.
Latest PMC-Sierra, Inc. Patents:
- Systems and methods for reclaiming memory for solid-state memory
- Systems and methods for adaptively selecting from among a plurality of error correction coding schemes in a flash drive for robustness and low latency
- Systems and methods for storing data in page stripes of a flash drive
- Scalable high-swing transmitter with rise and/or fall time mismatch compensation
- Systems and methods for decoding data for solid-state memory
This application is related to commonly-owned application titled RUN-LENGTH BASED SPECTRAL ANALYSIS, Ser. No. 12/055,948, filed on Mar. 26, 2008, now U.S. Pat. No. 8,019,028, issued on Sep. 13, 2011, the entirety of which is hereby incorporated by reference.
BACKGROUND1. Field of the Invention
The invention generally relates to electronics. In particular, the invention relates to filtering of interference.
2. Description of the Related Art
Co-existence of wireless communication links from different wireless standards, and a generally crowded wireless spectrum results in “interfering” radio signals near the frequency of a desired radio signal to be received, as illustrated in
In an extreme case, the presence of a relatively large interferer near the desired signal makes reception of the desired signal impossible. Even in a relatively good case, the ability to handle a relatively large interferer increases the linearity and baseband filtering requirements of the radio, which in turn increases the radio's cost and power.
One conventional solution to the problem of a large interferer is to increase the linearity and increase the analog baseband requirements of the radio front end. This approach increases both the cost and the power used by the radio.
In another approach illustrated in
Receivers for wireless radio typically tolerate interfering signals (interferers) in two basic duplex scenarios: Time Division Duplex (TDD) and Frequency Division Duplex (FDD).
In the illustrated example of
In a Frequency Division Duplex (FDD) system, a transceiver's own transmitter can be transmitting at the same time that it is receiving a signal. Due to the finite amount of attenuation of the local transmitter signal into the receive signal by the duplexer and other filtering, there may exist a residual transmit signal received with the received signal sufficient to cause distortion.
SUMMARY OF THE DISCLOSURELarge interfering signals (interferers) with spectra near a desired signal can cause distortion in a wireless receiver due to a non-linear signal path. It is typically a performance advantageous to attenuate these interferers earlier in the signal path, rather than later in the signal path, because these interferers can cause saturation of amplifying stages. In certain situations, the frequency offset of an interfering signal, with respect to the desired signal, can be on the order of 10 megahertz (MHz), whereas the center frequencies can be on the order of several gigahertz (GHz). Thus, a filter with “baseband” precision would be needed at radio frequency to notch out the interferer, which is relatively difficult to do. Disclosed is a technique to estimate the relative strength and center frequency of the interferer and to place the center frequency of a notch filter adaptively and precisely at the interferer location.
Embodiments of the invention preferably use an analog circuit with a relatively narrow notch characteristic to filter out the interferer at RF; estimate the strength and center frequencies of interferers; and control the center frequency and depth of the notch filter.
One embodiment is a calibration technique for locating a contour that is useful for control of a notch filter having a two-dimensionally control characteristic, such as a control for capacitance and a control for resistance. In the calibration technique, while holding a first control, such as a control for capacitance, constant, the second control, such as a control for resistance is varied in relatively widely spaced apart observations. A rough indication of a location of the notch is determined by these widely spaced apart observations. More closely spaced apart observations around the rough indication locate the location of the notch with greater precision. The procedure is repeated to collect data points indicating the location of the deepest notch in two dimensions. Then, a technique such as least-squares is used to fit the data to a contour, which should be used for filtering out interference.
These drawings and the associated description herein are provided to illustrate specific embodiments of the invention and are not intended to be limiting.
Although particular embodiments are described herein, other embodiments, including embodiments that do not provide all of the benefits and features set forth herein, will be apparent to those of ordinary skill in the art.
To adequately receive the desired signal in the presence of a large interferer as shown in
Choice 2 uses a relatively high precision center frequency and a relatively high Q notch filter. For example, the desired and interfering signals may be separated by as little as a few MHz. An example of a ratio of carrier frequency to center frequency or filter transition band is expressed in Equation 1.
Equation 1 illustrates that the ratio of filter frequencies to carrier frequency is relatively small. The high-Q nature of the filter may be managed using a resonating tank circuit, but the center frequency precision will typically be controlled with an active control loop. An applicable high-Q filter will be readily determined by one of ordinary skill in the art.
Typically, the active control loop of the high-Q filter will use an estimate of the center frequency of the interferer. Techniques to estimate the center frequency will now be described.
RF signals, including the desired signal and one or more interferers, are received by a low-noise amplifier (LNA) 406. The RF signals from the LNA 406 are converted to baseband by a downconverter 410.
An output of the downconverter 410 is provided to a slicer 412 and to other components 414, 416, 418, 420, 422 of the receiver front-end. In the illustrated embodiment, the slicer 412 samples the output of the downconverter 410 and determines whether the output is positive or negative. For example, the slicer 412 can generate hard symbols of zero or one from the output of the downconverter 410. The output of the slicer 412 is provided to the interference scanner 402. The interference scanner 402 will be described later in greater detail. The other components 414, 416, 418, 420, 422 can be arranged in a variety of ways, including, but not limited to, conventional ways. In one embodiment, one or more notch filters are incorporated into the LNA 406. Line 430 illustrates control of the one of more notch filters of the LNA 406 by the interference scanner 402.
With reference to
Consider two cases: one in which an interferer is 3.25*BWdes away from the desired signal, and another in which an interferer is 3.5*BWdes away, wherein BWdes is the bandwidth of the desired signal. For this example, the bandwidth BWdes=10 MHz, so the interferers are at 32.5 MHz and 35 MHz frequency offsets. The spectra of the two cases are shown in
A histogram of run-lengths from the output of the slicer 412 (
This data represented in the histogram raises 2 questions: (1) what is the relationship of run-length to interferer center frequency; and (2) run-lengths are discrete counts (natural number counts), but the interferer center frequency can be any frequency.
In one embodiment, equation 2 is used to convert a run-length to a signal frequency.
In Equation 2, Frunlength is the frequency of the interferer, Fsamp is the sampling frequency of the slicer 412 (
The run-lengths RL are of course discrete counts. For example, there cannot be a peak run length of 5.3 counts. The peak run length will be a discrete count, such as 5 or 6 counts in the illustrated example. However, data other than just the peak run length can also be used to evaluate a frequency of the interferer or a magnitude of the interferer. This other data is represented by the shape of the histogram. For example, points that are near the maximum frequency of occurrence run length can be used to estimate where the peak occurrence for run-length would have fallen if there had been a continuous run-length axis or a finer resolution count (faster sampling rate), that is, a non-natural number peak run-length. Techniques can also estimate where on the y-axis the maximum run-length would have fallen.
In the illustrate embodiment, the following Matlab® function can be used to estimate the continuous coordinates of the run-length with the maximum number of occurrences.
The above algorithm performs a linear extrapolation around the “raw” or discrete max to estimate an extrapolated max value. While the term extrapolation is used, the estimated data is within the run-length of the data (x-axis), but is outside the domain of the counted frequency of occurrence data (y-axis). The illustrated Matlab® function assumes that the peak is shaped like a simple “triangle” near the raw maximum (discrete count maximum). Visually, the algorithm can be observed in the graph of
For example, the points with run lengths 6 (maximum) and 7 (adjacent with lower count) are used for the curve that is extrapolated to a 32.5 MHz peak. For example, the points with run lengths 4 (maximum) and 5 (adjacent with lower count) are used for the curve that is extrapolated to a 35 MHz peak. The extrapolated peak is determined to be located at the intersection of said line and another line formed by negating the slope (changing the sign of the slope) of said line and passing said line through the nearest neighbor point that is closest to the maximum, such as the other adjacent point (point at run length 5 for the 32.5 MHz peak and the point at run length 6 for the 35 MHz peak). The foregoing illustrates that the extrapolated x-axis value (non-natural number run length) can be used to estimate a frequency of the interfering signal.
In addition, alternatively or in addition to the foregoing, an estimate of a signal strength of the interfering signal relative to a signal strength of the desired signal can be determined by examination of the magnitude of the extrapolated peak (y-axis). The estimated interfering signal strength can be used to determine whether to activate an interference filter, to assess the effectiveness of a particular interference filter configuration, to determine whether to adjust or tune an interference filter, or the like.
The foregoing algorithm can be implemented via hardware, firmware, software, or by a combination of the foregoing. For example, a microprocessor, microcontroller, or other processor can be used to assess the interferer frequency. Using such techniques, such as the foregoing algorithm, the coordinates of the peak of the interferer, which for the example of
The analysis of the run-lengths of the sign (positive or negative) of a signal can be used as a crude estimate of the spectrum of arbitrary signals, after the run-lengths are converted to frequencies, according to Equation 2. This analysis, illustrated with the aid of the histogram, should be limited to spectra with relatively few dominant peaks.
One application of the invention is in the field of wireless radio receivers; however, the interference scanner can be used for spectrum estimation for arbitrary signals.
In the illustrated example of
An output of the LNA 1104 is provided as an input to a downconverter 1110, which provides a downconverted signal as an input to baseband processing circuits 1112 and to an input of a notch filter control circuit. The illustrated notch filter control circuit includes slicer 1114, a processor 1116, a digital-to-analog converter 1118, and a notch filter circuit 1120. A circuit that can embody the LNA 1104 and the notch filter circuit 1120 will be described later in connection with
In the illustrated example, the capacitance and the negative resistance can be adjusted with DC voltage, VRV and VCV, respectively, which can be provided by outputs of the digital-to-analog converter 1118. One difficulty with the illustrated circuit is that both the center frequency (see
Values for the negative variable resistance RV are expressed along a horizontal axis. Values for the variable shunt capacitance CV are expressed along a vertical axis. Each curve in
As evidenced by the foregoing, the notch filter circuit uses a two-dimensional control: the variable negative resistor RV and the variable shunt capacitance CV. To place the notch filtering at a particular frequency, and to give it a relatively good, such as maximum depth, both the variable negative resistance RV and the variable shunt capacitance CV should be properly set.
One embodiment of the invention periodically updates an “R-C contour” which effectively stores parameters for an equation that generates RV and CV adjustment settings for a given desired notch center frequency. As long as the parameters for the equation are fresh, a notch can be placed virtually instantly at the location of the interferer.
RV and CV can be modeled as a polynomial in “fdesired” or “f”, the desired notch frequency, as shown in Equations 3 and 4.
Rv(fdesired)=aRv1f+aRv2f2+ . . . Equation 3
Cv(fdesired)=aCv1f+aCv2f2+ . . . Equation 4
An overview of a technique that can be used to find settings for the variable negative resistor RV and the variable shunt capacitance CV is set forth below. With reference to
In one embodiment, rather than counting run lengths, the estimated spectrum 1604 is obtained by averaging, in the frequency domain, N M-point FFTs (fast Fourier transforms) of a sign (zero or one) of the down-converted signal, wherein in the illustrated example, M=32, and N=128. Other values for M and N are applicable. As illustrated by
With particular settings for the variable negative resistor RV and the variable shunt capacitance CV, the center frequency of the notch and the depth of the notch should vary. The notch filter circuit 1120 notches out a corresponding portion of the frequency spectrum, and the frequency spectrum is analyzed by the slicer 1114 and the processor 1116.
In one embodiment, a broadband noise source is enabled, a setting is selected for the variable negative resistor RV and for the variable shunt capacitance CV, and the characteristics of the notch are analyzed. Various other settings for the variable negative resistor RV and for the variable shunt capacitance CV are selected and analyzed to provide data points for establishing the curves and contours described earlier in connection with
In the illustrated embodiment, the contour 1404 is quickly and efficiently located by first using a relatively broad data spacing and then using relatively tight data spacing, as will be discussed below. The technique finds the contour 1404 quickly, which can also save battery power. However, in an alternative embodiment, a brute force technique utilizing an exhaustive search of data can be used.
In the illustrated embodiment, the estimated spectrum 1604 is observed first by holding the variable shunt capacitance CV to a constant amount of capacitance, and varying the variable negative resistance RV. In a first control loop, the variable shunt capacitance CV is set to about 1.35 picofarads (pF), and the variable negative resistance is varied from 215 ohms to 260 ohms (that is, resistance is −215 ohms to −260 ohms), in relatively broad steps, such as 5 ohm steps. It will be understood that the processor 1116 does not need to be aware of a particular voltage setting for a particular amount of capacitance, such as 1.35 pF, because the notch filtering can be configured adaptively. However, if desired, the voltage settings for particular ranges of capacitance can be stored in a lookup table.
For each of the variable negative resistance RV values, an estimated spectrum, such as the estimated spectrum 1604, is observed. The observation points are illustrated by “dots” 1702 in
A second set of data points is then taken around the point with the deepest notch. Thus, for a shunt capacitance of 1.35 pF, more data points (tighter spacing) 1704 are taken from about 224 ohms to 234 ohms to assist in locating the point along the contour 1404 with more precision. For example, 10 data points are taken from about 224 ohms to 234 ohms in 1 ohm intervals. In the illustrated embodiment, a point at 229 ohms is found to contain the deepest depth and is highlighted in
The shunt capacitance CV is then adjusted, and another data point having the deepest depth is found and so on. In one embodiment, the process assumes a slope, based on, for example, simulation data, bench testing, production testing or the like, in the Negative R, CV plane for a contour having the deepest depth, and only “fine” or tight spacing data points are collected for finding the data point having the deepest depth. In an alternative embodiment, the data collection techniques with an initially coarse spacing and a subsequent narrow spacing are repeated. The resulting data points for the narrow spacings and the larger dots for the determined notch position are illustrated in
The collection of points with the deepest depth, for example, the large dots 1706, are analyzed using a technique such as least squares, and the coefficients to Equations 1 and 2 are derived to model the contour 1404 for the maximum depth. The least squares fit is illustrated as a contour 1708 in
In an alternative embodiment, the order of adjusting the shunt capacitance CV and the negative resistance RV is interchanged. For example, rather than holding the shunt capacitance CV constant while varying the negative resistance RV, and then adjusting the shunt capacitance CV and repeating, the negative resistance RV is held constant, then the shunt capacitance CV is varied while the notch filter circuit 1120 is analyzed.
The process can start at a wait state 2002, where the process generally resides unless a particular activity is selected. The three basic activities performed by the illustrated process include (1) updating the adjustable negative resistance and adjustable capacitance (RV-CV) table or equations (such as Equations 1 and 2); (2) updating a center frequency of the notch; and (3) estimating an interferer strength.
When in a calibration mode, which can be activated when the corresponding device is not receiving data, the tables or parameters for the equations (Equations 1 and 2) can be updated 2004. After the table or parameters are updated, the process can return to the wait state 2002.
Preferably, periodically, the process activates the state 2006 to estimate the interferer strength. For example, the techniques described in U.S. patent application Ser. No. 12/055,948 can be used. The process then determines whether the strength is greater than or equal to a threshold (THR) or is less than the threshold. The threshold should indicate a level at or above which, the interferer is deemed to be sufficiently strong that it should be filtered, whereas below this threshold, the interferer does not have to be filtered.
If the interferer strength is estimated to be above the threshold, then the process advances from the state 2006 to a state 2008, in which the interferer's frequency is estimated. The interferer's frequency is analyzed to determine whether it had previously been encountered (that is, is “old”) or is a new interferer. In one embodiment, estimated frequencies near to each other are considered to be from the same interferer, that is, the same interferer may be detected at slightly different frequencies.
If the process determines that an “old” interferer is being detected, then the process advances from the state 2008 to the state 2010, where the process combines previous frequency measurement data, such as prior run-length data, with the currently measured data to help refine the frequency estimate. The process then advances from the state 2010 to a state 2012, and updates the interferer strength in a list of detected interferers. The list implements a peak detector, so that an interferer that has been filtered out by the notch filter circuit 1120 can remain filtered. The detected interferers are identified by frequency and by an estimated interferer strength, as obtained earlier in the state 2006. In one embodiment, the list is maintained in table form. This allows the process to compare the relative strengths of each interferer.
The process then advances from the state 2012 to the state 2014, which can update the decision of which interferer(s) to filter. For example, there can be more interferers than notch filter circuits, and typically, the strongest interferers are the ones that should be filtered.
Returning now to the state 2006, when the interferer is estimated to be below the threshold for filtering, the process advances to the state 2012, which was previously described.
Returning now to the state 2008, when the interferer is deemed to be “new,” that is, not previously encountered, then the interferer frequency list is updated 2016, that is, an entry for the new interferer is added to the interferer frequency list. The process then advances to the state 2012, which was previously described.
The list maintained by the states 2016 and 2012 permits an interferer that has been effectively removed by the notch filtering process to remain in the list to be considered for filtering, and this prevents the process from engaging filters on and off for a single interferer in a rapid oscillation.
In one embodiment, when a receiver of a transceiver is not receiving data, the notch filtering for an interferer can be disengaged to determine whether that interferer is still present. This refreshes the peak level for the interferer's frequency. If the interferer is no longer present, it can be removed from the interferer frequency list. In an alternative embodiment, the depth of filtering for a particular interferer can be diminished over time to gracefully determine whether the interferer remains present.
Various embodiments have been described above. Although described with reference to these specific embodiments, the descriptions are intended to be illustrative and are not intended to be limiting. Various modifications and applications may occur to those skilled in the art.
Claims
1. An apparatus comprising:
- a low noise amplifier (LNA) configured to receive a radio frequency input signal, wherein the LNA is configured to generate an amplified RF signal;
- a downconverter operatively coupled to the LNA to receive the amplified RF signal, wherein the downconverter is configured to generate a baseband signal as an output;
- a slicer operatively coupled to an output of the downconverter, wherein the slicer is configured to generate hard-decision samples of ones and zeroes as an output;
- a processor operatively coupled to the slicer, wherein the processor is configured to analyze the hard-decision samples to generate a control signal for control of filtering of at least one frequency range of the amplified RF signal; and
- a controllable notch filter operatively coupled to an output of the processor and to an output of the LNA, wherein the controllable notch filter is configured to receive the control signal and to filter out the at least one frequency range from the amplified RF signal.
2. The apparatus of claim 1, wherein the processor is further configured to compute Fourier Transforms of the hard-decision samples to generate an estimate of a frequency of an interferer or of the at least one filter frequency of the controllable notch filter.
3. The apparatus of claim 2, wherein the processor is further configured to compute a plurality of fast Fourier Transforms, and to generate an average of the plurality of fast Fourier Transforms to generate the estimate of the frequency of the interferer or the at least one filter frequency of the controllable notch filter.
4. The apparatus of claim 3, wherein the processor is further configured to compute 128 32-point fast Fourier Transforms, and to generate an average of the 128 32-point fast Fourier Transforms to generate the estimate of the frequency of the interferer or the at least one filter frequency of the controllable notch filter.
5. The apparatus of claim 2, further comprising a broadband noise source operatively coupled to an input of the LNA, wherein the broadband noise source is activated such that the processor can generate the estimate of the at least one filter frequency of the controllable notch filter.
6. The apparatus of claim 2, wherein the LNA is configurable to be able to switch to a high-noise state to generate broadband noise for spectral analysis of the controllable notch filter.
7. The apparatus of claim 1, wherein the processor further comprises a digital-to-analog converter configured to convert the control signal from digital form to analog form, and to provide the control signal to the controllable notch filter in analog form.
8. The apparatus of claim 1, wherein the processor is configured to identify one or more interferers from analysis of the hard decision samples, to store previously detected interferer levels in a list, and to determine the at least one frequency to filter out based at least partly on the previously detected interferer levels.
9. The apparatus of claim 1, wherein the processor is configured to identify one or more interferers from analysis of the hard decision samples, to store previously detected interferer levels in a list, to determine whether a currently detected interferer had been previously detected, and to combine current data with prior data to refine a center frequency estimate of the currently detected interferer.
10. The apparatus of claim 1, wherein the processor is configured to identify one or more interferers from analysis of the hard decision samples, is configured to temporarily disable the controllable notch filter to refresh a peak level of a previously detected interferer frequency, and wherein the processor is further configured to reevaluate whether to continue to filter out the previously identified and filtered interferer based at least partly on the refreshed peak level.
11. The apparatus of claim 1, wherein the processor is configured to identify one or more interferers from analysis of the hard decision samples, to gradually change a depth of filtering of the controllable notch filter to determine whether a previously identified and filtered interferer remains present, and wherein the processor is configured to reevaluate whether to continue to filter out the previously identified and filtered interferer based at least partly on the determination of presence.
12. The apparatus of claim 1, wherein a frequency and depth of the controllable notch filter is two-dimensionally controlled, wherein the processor is configured to generate a two-dimensional control for the control signal, wherein both a frequency and depth of filtering is two-dimensionally controlled such that both frequency and depth of filtering varies in two dimensions.
13. A method of filtering in a radio frequency front-end, the method comprising:
- receiving a radio frequency input signal and amplifying the radio frequency input signal to generate an amplified RF signal;
- downconverting the amplified RF signal to generate a baseband signal;
- generating hard-decision samples of ones and zeroes from the baseband signal;
- analyzing the hard-decision samples to detect at least one interferer and to generate a control signal for control of filtering of at least one frequency range of the amplified RF signal corresponding to the at least one interferer; and
- notch filtering out the at least one frequency range from the amplified RF signal based at least partly on the control signal to filter out the at least one interferer.
14. The method of claim 13, further comprising maintaining a list of the one or more interferers, and determining the at least one frequency to filter out based at least partly on the list.
15. The method of claim 13, further comprising maintaining a list of the one or more interferers, temporarily disabling notch filtering to refresh a peak level of a previously detected interferer frequency, and reevaluating whether to continue to filter out the previously detected and filtered interferer based at least partly on the refreshed peak level.
16. The method of claim 13, further comprising:
- maintaining a list of the one or more interferers;
- determining whether a currently detected interferer had been previously detected;
- combining current data with prior data to refine a center frequency estimate of the currently detected interferer.
17. The method of claim 13, further comprising gradually changing a depth of filtering to determine whether a previously detected and filtered interferer remains present, and reevaluating whether to continue to filter out the previously detected and filtered interferer based at least partly on the determination of presence.
18. A method of filtering an interferer in a radio frequency front-end, the method comprising:
- receiving a radio frequency input signal and amplifying the radio frequency input signal to generate an amplified RF signal;
- downconverting the amplified RF signal to generate a baseband signal;
- generating hard-decision samples of ones and zeroes from the baseband signal;
- analyzing the hard-decision samples to generate a control signal for control of filtering of at least one frequency range of the amplified RF signal;
- filtering out the at least one frequency range from the amplified RF signal based at least partly on the control signal; and
- computing Fourier Transforms of the hard-decision samples to generate an estimate of a frequency of the interferer or the at least one filter frequency of filtering.
19. The method of claim 18, further comprising computing a plurality of fast Fourier Transforms, and generating an average of the plurality of fast Fourier Transforms to generate the estimate of the frequency of the interferer or filtering.
20. The method of claim 19, further comprising computing 128 32-point fast Fourier Transforms, and generating an average from the 128 32-point fast Fourier Transforms to generate the estimate of the frequency of the interferer or filtering.
21. The method of claim 18, further comprising activating a broadband noise source for estimation of the at least one filter frequency.
22. The method of claim 18, further comprising configuring a low-noise amplifier (LNA) of the radio frequency front-end to switch to a high-noise state to generate broadband noise for estimation of the at least one filter frequency.
23. The method of claim 13, further comprising generating a two-dimensional control for the control signal, wherein both a frequency and depth of the notch filtering is two-dimensionally controlled such that both frequency and depth of filtering varies in two dimensions.
24. A method of locating a contour of optimal attenuation for a notch filter having a two-dimensional control characteristic so that both frequency and depth of filtering vary according to variation with a first control input and a second control input, the method comprising:
- (a) activating broadband noise in a front-end of a wireless receiver for calibration;
- (b) filtering the broadband noise in the front-end of the wireless receiver with the notch filter;
- (c) selecting an initial settings for the first control input of the notch filter, and then performing (d)-(f): (d) observing a frequency response of the notch filter for a first plurality of settings of the second control input, wherein the first plurality of settings cover a first range and are spaced apart at a first spacing; (e) observing the frequency response of the notch filter for a second plurality of settings of the second control input, wherein the second plurality of settings cover a second range that is smaller than the first range and covers a frequency response observation from the first range that has the deepest notch characteristic, and wherein the second plurality of settings are spaced apart by a second spacing that is tighter than the first spacing; (f) collecting the setting from within the second plurality of setting having the deepest notch characteristic for the particular settings of the first control input and the second control input;
- (g) repeating (d)-(f) for other settings of the first control input; and
- (h) determining a contour for control of the notch filter based on the collected setting for the first control input and the second control input having the deepest notch characteristic.
25. The method of claim 24, wherein observing the frequency response comprises estimating the frequency response by computing a fast Fourier transform of a sign of a downconverted signal associated with the broadband noise as filtered by the notch filter.
26. The method of claim 24, further comprising performing (d), (e), and (f) in sequence.
27. The method of claim 24, wherein the first control input corresponds to capacitance control and the second control input corresponds to resistance control.
28. The method of claim 24, wherein the second spacing is tighter than the first spacing by at least a factor of 4.
29. The method of claim 24, wherein determining the contour comprises using a least squares fit to the collected deepest notch characteristic settings to find coefficients that model the contour.
30. An apparatus comprising:
- (a) a controllable source of broadband noise;
- (b) a notch filter having a two-dimensional control characteristic so that both frequency and depth of filtering vary according to variation with a first control input and a second control input; and
- (c) a processor configured to control activation of the controllable source of broadband noise and control filtering by the notch filter, the processor configured to select an initial settings for the first control input of the notch filter, and then is configured to (d)-(f): (d) observe a frequency response of the notch filter for a first plurality of settings of the second control input, wherein the first plurality of settings cover a first range and are spaced apart at a first spacing; (e) observe the frequency response of the notch filter for a second plurality of settings of the second control input, wherein the second plurality of settings cover a second range that is smaller than the first range and covers a frequency response observation from the first range that has the deepest notch characteristic, and wherein the second plurality of settings are spaced apart by a second spacing that is tighter than the first spacing; (f) collect the setting from within the second plurality of setting having the deepest notch characteristic for the particular settings of the first control input and the second control input;
- (g) wherein the processor is configured to repeat (d)-(f) for other settings of the first control input;
- (h) wherein the processor is configured to determine a contour for control of the notch filter based on the collected setting for the first control input and the second control input having the deepest notch characteristic.
31. The apparatus of claim 30, wherein the processor is configured to observe the frequency response by estimation of the frequency response via computation of a fast Fourier transform of a sign of a downconverted signal associated with the broadband noise as filtered by the notch filter.
32. The apparatus of claim 30, wherein the processor is configured to perform (d), (e), and (f) in sequence.
33. The apparatus of claim 30, wherein the first control input corresponds to capacitance control and the second control input corresponds to resistance control.
34. The apparatus of claim 30, wherein the second spacing is tighter than the first spacing by at least a factor of 4.
35. The apparatus of claim 30, wherein the processor is configured to determine the contour via computation of a least squares fit to the collected deepest notch characteristic settings to find coefficients that model the contour.
Type: Grant
Filed: Jan 6, 2009
Date of Patent: Apr 23, 2013
Assignee: PMC-Sierra, Inc. (Sunnyvale, CA)
Inventors: Anthony Eugene Zortea (Pipersville, PA), Mathew McAdam (Vancouver), Mathieu Gagnon (Verdun), Graeme Boyd (North Vancouver), Winston Ki-Cheong Mok (Vancouver)
Primary Examiner: Erin File
Application Number: 12/349,328
International Classification: H03D 1/04 (20060101); H03D 1/06 (20060101); H03K 5/01 (20060101); H03K 6/04 (20060101); H04B 1/10 (20060101); H04L 1/00 (20060101); H04L 25/08 (20060101);